经济学人:人脸识别技术
首页 / 英语, 英国 / 经济学人广播The Economist / 播客 /
发布于: 5年前 5'57" 143wpm
In 2018, machines that can read your face will go mainstream, changing the way we live. Your face will become your password, unlocking smartphones and bank accounts.
“Face recognition in 2018 will change our lives.”
But the technology will also have the power to covertly track your movements.
“Moving into a world where machines can follow you around and characterise you based only on your face is worrying because it potentially limits your freedom.”
It will even be able to guess your sexuality through facial features alone.
In 2018, we’ll be forced to face the future.
The human face has an astonishing variety of features, which not only help us recognise others, but read and understand them through a constant flow of intentional and unintentional signals. It’s one of the unique functions that separates man from machine, until now.
“2018 is going to be a landmark year for face recognition because there’s been a huge amount of progress in the underlying technology known as machine learning, and what that allows you to do is pull a very accurate faceprint out of a photograph that uniquely identifies a person.”
Pioneering facial-recognition technology hasn’t yet hit the mainstream. In 2018, it will be in our pockets.
“Facial recognition is going to go mainstream in 2018 largely because the latest iPhone has face-recognition technology built into it.”
“You look at it and it recognises you.”
“By next year, there’ll be millions of those devices out in the world and we’re going to get used to unlocking our phone, then we’re going to start getting used to getting into our banking apps with face recognition, and it’s going to start feeling normal.”
But using your face to unlock your phone is just the beginning. In the suburbs of Israel’s financial center Tel Aviv, a team of engineers is at the forefront of a technological revolution.
“We have invented a simple way of transforming a photo into text.”
They’re teaching machines to read faces.
“The face-recognition model can measure the distance between your eyes, the width of your lips, the distance from your lips to your nose, etc. At the end of that process, we are left with a simple text, which is the basis of our recognition.”
The software has the power to identify one face from millions in under one second.
“Our accuracy under a controlled environment is around 99%.”
And it’s this precision that makes the technology an effective new tool for surveillance.
“We can identify people without them knowing that they are being identified. We have clients ranging from medical centers, retail chain stores, churches.”
Retail stores are using this technology to generate data on customers, tracking their shopping habits and targeting in-store adverts. Churches are even using facial recognition to monitor attendance. And one school in the UK wants to use it to keep tabs on teachers.
“We have one school where teachers took too long breaks. Therefore, the management decided to put them on the watch-list.”
But there’s one country that’s ahead of the game.
“China is a harbinger of the future for this kind of technology.”
Companies have access to a government image database of 700 million people – half of its population.
“China is already using this widely in security, looking for terrorists or people who have warrants out for their arrest. But they’re also using it to let people pay in fast-food restaurants, to access theme parks without having to buy a ticket, and even, bizarrely, to try and catch people who steal loo paper from public toilets.”
But there’s the potential for more sinister applications. Researchers have shown that your face can point to your sexuality.
“It’s not going to mean you can take a photo of someone and get a yes or no, gay or straight. But what it does mean is that if a bad actor or an authoritarian regime gathers, say, 10,000 pictures and asks which of the 10 people are most likely to be gay, they’re quite likely to get many of those right.”
This new ability to record, store and analyze images of faces on a vast scale will fundamentally change notions of privacy, fairness and trust.
“In 2018, the most contact that people will have with face recognition they won’t actually know necessarily, it’ll be through security services, it’ll be through shops that are profiling you and that’s the more worrying side of it.”
But tech companies are forging ahead with their plans to make facial recognition an everyday part of our lives.
“You cannot stop progress on [over] the fear of privacy. The advantages of security, of business information are far greater than some privacy issues.”
“This technology is a new kind of power and the only way we have of dealing with technologies currently is to make regulations that repress bad uses and allow for the perceived good uses. 2018 will be a year we start to work out what some of those are.”
相关播客
Comment...